Power grid line loss intelligent diagnosis and visualization method based on multi-source data acquisition

Abstract

The invention relates to the technical field of three-dimensional visualization, in particular to a power grid line loss intelligent diagnosis and visualization method based on multi-source data acquisition, and the method comprises the steps: obtaining a sensor data grouping sequence, detecting power fluctuation synchronism, extracting a change trend, analyzing a power difference and a node trend, marking an abnormal path, generating a line loss map, and positioning an abnormal node coordinate level. And mapping the topological graph to adjust the style, and generating a visual layer result. According to the invention, through obtaining the multi-dimensional data of the feeder line and performing efficient analysis, capturing power fluctuation and abnormity, identifying abnormal nodes in combination with the power difference of adjacent nodes, analyzing the fluctuation rate to reveal the line loss area and the structure abnormity, calibrating the abnormal position by using the three-dimensional coordinate, realizing spatial positioning and risk assessment, and clearly displaying the operation state by the three-dimensional layer. The monitoring accuracy is improved, path connection and node positioning help to quickly identify a high-risk area, the operation and maintenance efficiency and the diagnosis comprehensiveness are optimized, and a precise basis is provided for power grid management.

Description

Technical Field

[0001] The present invention relates to the field of three-dimensional visualization technology, and in particular to a method for intelligent diagnosis and visualization of power grid line losses based on multi-source data acquisition. Background Art

[0002] The field of three-dimensional visualization technology includes technical directions such as spatial data modeling, three-dimensional scene construction and dynamic interactive presentation. This field transforms abstract data into visual entities with spatial dimensions by establishing data mapping relationships under a three-dimensional coordinate system. Combined with lighting rendering, perspective transformation and interactive operation technologies, it realizes the three-dimensional spatial representation of the operating status of complex systems. In power grid monitoring, it is mainly used for displaying equipment topology structures, tracking energy flow paths and spatial positioning of abnormal states.

[0003] The intelligent diagnosis and visualization method for power line losses based on multi-source data acquisition constructs a three-dimensional topological mapping model of power line loss characteristics by fusing multidimensional data from distribution transformers, smart meters, and line loss monitoring devices. This method uses a data fusion engine to correlate the multidimensional characteristics of current and voltage phase angle differences, power factor fluctuations, and temperature and humidity environmental parameters. A line loss gradient shading algorithm in a three-dimensional spatial coordinate system is used to create a spatial distribution heat map of line loss values. Dynamic topological node calibration technology is used to locate abnormal line loss areas in three dimensions, ultimately creating a layered, rendered, three-dimensional visualization of power line loss diagnostics.

[0004] Existing technologies in power grid line loss monitoring mainly rely on multi-source data integration and simple topology display, which can perform preliminary monitoring. However, they have limitations when faced with complex power grid topologies and large amounts of data. They lack in-depth identification of power fluctuations in power feeders and node differences, and fail to effectively explore the detailed changes in the relationship between power fluctuations and nodes, resulting in difficulty in accurately locating abnormal nodes. In addition, there is a lack of line loss positioning methods based on three-dimensional spatial coordinates, and it is impossible to intuitively display the associations and abnormal states between power grid nodes. The monitoring results are usually static topology maps, lacking spatial hierarchy and operability. The technology fails to respond to changes in the power grid structure in a timely manner, resulting in difficulty in timely detection of abnormal areas in sudden line losses, affecting the speed of problem solving. Existing technologies are insufficient in their ability to cope with complex power grid environments and lack sufficient visualization and interactivity, affecting the accuracy and timeliness of diagnosis. Summary of the Invention

[0005] In order to solve the technical problems existing in the prior art, the embodiment of the present invention provides a method for intelligent diagnosis and visualization of power grid line losses based on multi-source data collection. The technical solution is as follows:

[0006] The intelligent diagnosis and visualization method of power grid line loss based on multi-source data collection includes the following steps:

[0007] S1: Obtain the sensor data of each feeder, group them by number and arrange them in time sequence, detect the synchronization of power fluctuations, extract the change trend, and obtain the feeder power fluctuation sequence results;

[0008] S2: Call the feeder power fluctuation sequence results, analyze the difference between the power input and output of adjacent feeder nodes, compare the power difference with the node sequence number change trend, mark the nodes with inconsistent fluctuation trends as path anomalies, and obtain the feeder line loss mutation marking results;

[0009] S3: extracting abnormal nodes in the feeder line loss mutation mark result, calculating the power change rate difference, extracting the alternating fluctuation node group, forming a path closure graph, and generating a line loss structure abnormal path graph result;

[0010] S4: extracting the topological number and position code of the abnormal node in the abnormal path map of the line loss structure, attaching the node to the spatial coordinate, marking the hierarchical distribution, outputting the position and hierarchical information, and obtaining the three-dimensional structure risk positioning information result;

[0011] S5: Based on the three-dimensional structure risk location information results, map abnormal nodes to the topological structure diagram, adjust the rendering style, determine the layer identification based on the spatial position, complete the visual expression of nodes and paths, and generate the power grid line loss visualization layer result.

[0012] As a further solution of the present invention, the feeder power fluctuation sequence results include voltage data, current data, power factor data, conductor temperature rise data and loop length data; the feeder line loss mutation marking results include adjacent feeder node power input values, adjacent feeder node power output values, power difference change trends, node sequence number change trends and path abnormal nodes; the line loss structure abnormal path map results include path abnormal node sets, low-voltage access points, terminal metering equipment, branch buses and path node connection sequences; the three-dimensional structure risk positioning information results include feeder topology numbers, distribution level identifications, geographic location codes, spatial coordinate positions and node level information; the power grid line loss visualization layer results include feeder path topology diagrams, terminal load unit distribution diagrams, branch structure level diagrams, node and path visual expressions and layer identification forms.

[0013] As a further solution of the present invention, the steps for obtaining the feeder power fluctuation sequence result are:

[0014] S101: Obtain the voltage, current, power factor, conductor temperature rise, and loop length sensor data corresponding to each feeder number of the distribution unit, establish a data group index based on the feeder number, and horizontally splice different sensor data with the same number based on the acquisition timestamp to generate a feeder time series data set;

[0015] S102: Based on the feeder time series data set, arrange each feeder data group in ascending order by timestamp, extract power values ​​at adjacent time points to form a difference sequence, set a power fluctuation threshold range, identify the starting time points of sections that continuously exceed the threshold, and generate a synchronous fluctuation time mark group;

[0016] S103: Call the synchronous fluctuation time mark group, intercept the power data sequence of the corresponding time segment of each feeder, calculate the linear regression slope of the power value in each time segment, and connect the segments whose absolute value of the slope exceeds the set reference value in series in chronological order to generate the feeder power fluctuation sequence result.

[0017] As a further solution of the present invention, the step of obtaining the feeder line loss mutation marking result is:

[0018] S201: Calling the power input value and output value sequence of adjacent feeder nodes in the feeder power fluctuation sequence result, calculating the input and output differences by aligning the time points, and counting the absolute value of the power difference of each feeder node pair to generate a feeder power difference sequence;

[0019] S202: extracting a moving average of difference values ​​in a continuous time window in the feeder power difference sequence, calculating a Pearson correlation coefficient between an increasing trend of sequence numbers of adjacent feeder nodes and a changing trend of difference values, and generating a trend difference identifier;

[0020] S203: Filtering node pairs whose Pearson correlation coefficients are lower than a set threshold according to the trend difference identifier, marking the node with a larger sequence number in the node pair as a path abnormal node, and generating a feeder line loss mutation mark.

[0021] As a further solution of the present invention, the steps for obtaining the abnormal path map result of the line loss structure are:

[0022] S301: Calling the path abnormal node set in the feeder line loss mutation mark, matching the low-voltage access point coordinates, terminal metering equipment numbers and branch bus connection relationships in the distribution network topology structure, and generating an abnormal node topology association table;

[0023] S302: Based on the abnormal node topology association table, extract the power values ​​of each abnormal node for three consecutive time periods, calculate the absolute value of the difference between the power change rates of adjacent periods, and generate the node power fluctuation gradient;

[0024] S303: Based on the node power fluctuation gradient, identify the node group whose fluctuation direction reverses twice in succession, construct the conduction path between the nodes according to the connection order of the distribution network, and generate a line loss structure abnormal path map.

[0025] As a further solution of the present invention, the steps for obtaining the three-dimensional structure risk positioning information result are:

[0026] S401: Calling the abnormal path nodes in the abnormal line loss structure path map, extracting the corresponding feeder topology number and distribution level identifier, matching the geographic coordinate code in the power grid GIS system, and generating an abnormal node spatial attribute table;

[0027] S402: Based on the abnormal node spatial attribute table, the hierarchical structure is divided according to the feeder voltage level, the node geographic coordinates are converted into X / Y / Z axis values ​​in a three-dimensional coordinate system, and a node spatial coordinate mapping set is generated;

[0028] S403: According to the node space coordinate mapping set, a vector relationship of the connection path between nodes is established in a three-dimensional coordinate system, and the distribution level identifier corresponding to each node is superimposed to generate three-dimensional structural risk positioning information.

[0029] As a further solution of the present invention, the steps for obtaining the grid line loss visualization layer result are:

[0030] S501: Calling the node coordinates and level identifiers in the three-dimensional structure risk location information, matching the corresponding node positions in the feeder path topology structure diagram, setting the red dotted line channel rendering style according to the path connection status, and generating an abnormal path rendering base map;

[0031] S502: Rendering a base map based on the abnormal path, superimposing the equipment point data of the terminal load unit distribution map, assigning layer identifiers of different shapes according to the spatial positions of the nodes, and generating a composite spatial identification layer;

[0032] S503: Integrate the hierarchical division data of the composite space identification layer and the branch structure level map, uniformly adjust the transparency and superposition order of each layer, and generate a power grid line loss visualization layer.

[0033] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0034] In the present invention, by acquiring multi-dimensional data of the power grid feeder and efficiently integrating and analyzing it, the power fluctuations and abnormal changes of the power grid system are accurately captured, the power change trend of the feeder is monitored in real time, and the abnormal nodes are identified in combination with the power differences of adjacent nodes. The power fluctuation rate of the node is further analyzed to reveal the potential areas and structural abnormalities of the power grid line loss. The innovation lies in associating nodes with feeder paths and accurately calibrating abnormal areas using a three-dimensional spatial coordinate system to achieve spatial positioning and risk assessment of line losses. The three-dimensional layer structure display makes the power grid operation status and abnormal areas intuitive and clear, improving the efficiency and accuracy of monitoring and maintenance. Accurate node positioning and path connection help power grid managers quickly locate potential high-risk areas, take timely measures to reduce losses and optimize operation and maintenance efficiency, and optimize the overall processing logic optimization to improve the flexibility and comprehensiveness of power grid line loss diagnosis, providing an accurate basis for power grid operation and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a flow chart of the method of the present invention;

[0036] Figure 2 A flow chart for obtaining feeder power fluctuation sequence results of the present invention;

[0037] Figure 3 A flow chart for obtaining feeder line loss mutation marking results according to the present invention;

[0038] Figure 4 A flow chart for obtaining the abnormal path map results of the line loss structure of the present invention;

[0039] Figure 5 A flowchart for obtaining the three-dimensional structure risk positioning information results of the present invention;

[0040] Figure 6 This is a flowchart for obtaining the grid line loss visualization layer results of the present invention. DETAILED DESCRIPTION

[0041] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0042] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0043] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same. The terms "of," "corresponding," and "corresponding" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same.

[0044] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0045] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0046] See also Figure 1The present invention provides a technical solution: a method for intelligent diagnosis and visualization of power grid line losses based on multi-source data acquisition, comprising the following steps:

[0047] S1: Obtain the voltage, current, power factor, conductor temperature rise, and loop length sensor data of each feeder in the distribution unit, group them by feeder number, splice the data in the same group and arrange them in chronological order, detect the synchronization of power fluctuation sections, extract the power change trend of continuous sections, and obtain the feeder power fluctuation sequence results;

[0048] S2: Call the power input and output value sequences of each adjacent feeder node in the feeder power fluctuation sequence results, calculate the power input and output differences of the adjacent nodes respectively, compare the power difference change trends, check the correspondence between the node sequence number change trends and the power difference change trends, determine whether the fluctuation trends are consistent, mark the nodes with inconsistent change trends as path abnormal nodes, and obtain the feeder line loss mutation marking results;

[0049] S3: Extract the set of path abnormal nodes from the feeder line loss mutation marking results, locate the low-voltage access points, terminal metering equipment, and branch busbars associated with the feeder path nodes, calculate the difference in power change rate of the path abnormal nodes based on the power change trend of the path abnormal nodes, compare the power change trend and power change rate relationship before and after the path abnormal nodes, extract the path node groups with alternating fluctuations in the change trend, form a closed path graph of the path node connection sequence, and generate the line loss structure abnormal path graph result;

[0050] S4: Extract the feeder topology number, distribution level identifier, and geographic location code of the abnormal node in the line loss structure abnormal path map, and classify the levels according to the feeder structure. By attaching the nodes to the spatial coordinate system, marking the hierarchical distribution position corresponding to the path association relationship, and outputting the node position and level information in the spatial coordinate system, the three-dimensional structure risk positioning information result is obtained;

[0051] S5: Based on the three-dimensional structure risk location information results, the abnormal path nodes are mapped to the feeder path topology diagram, the terminal load unit distribution diagram, and the branch structure level diagram. The channel rendering style is adjusted according to the path connection status in the diagram, and the layer identification form is determined based on the spatial position of the node. The visual expression of the nodes and paths in the three-dimensional structure is completed, and the grid line loss visualization layer results are generated.

[0052] The feeder power fluctuation sequence results include voltage data, current data, power factor data, conductor temperature rise data and loop length data. The feeder line loss mutation marking results include the power input value of adjacent feeder nodes, the power output value of adjacent feeder nodes, the power difference change trend, the node sequence change trend and the path abnormal nodes. The line loss structure abnormal path map results include the path abnormal node set, low-voltage access point, terminal metering equipment, branch bus and path node connection sequence. The three-dimensional structure risk positioning information results include feeder topology number, distribution level identification, geographic location code, spatial coordinate position and node level information. The power grid line loss visualization layer results include feeder path topology structure diagram, terminal load unit distribution diagram, branch structure level diagram, node and path visual expression and layer identification form.

[0053] See also Figure 2 , the steps to obtain the feeder power fluctuation sequence results are:

[0054] S101: Obtain the voltage, current, power factor, conductor temperature rise, and loop length sensor data corresponding to each feeder number of the distribution unit, establish a data group index based on the feeder number, and horizontally splice different sensor data with the same number based on the acquisition timestamp to generate a feeder time series data set;

[0055] The sensor data of two feeders numbered F01 and F02 at two consecutive time points (T1 = 2025-07-04 10:00:00, T2 = 2025-07-04 10:00:05) are obtained from each data acquisition terminal of the power distribution unit. The AC voltage effective value of feeder F01 at time T1 is 220.1V, the current effective value is 15.2A, the power factor is 0.98, and the difference between the conductor temperature and the ambient temperature is 25.1°C. The physical length of the loop is 15.5 meters. The corresponding values ​​at time T2 are 220.3V, 15.8A, 0.98, 25.9℃, and 15.5 meters. The corresponding values ​​of feeder F02 at time T1 are 219.9V, 18.1A, 0.99, 28.3℃, and 21.8 meters. The corresponding values ​​at time T2 are 220.0V, 18.3A, 0.99, 28.6℃, and 21.8 meters. A data index structure with the feeder number as the unique key value is established. Specifically, an entry is created with the key string 'F01', whose value points to a data set. Another entry is then created for feeder 'F02'. Five different types of sensor data, including voltage, current, power factor, conductor temperature rise, and loop length, collected from the same feeder number are associated and merged horizontally based on their shared acquisition timestamps. For example, for feeder F01, its five data values ​​at timestamp T1 (220.1V, 15.2A, 0.98, 25.1°C, and 15.5m) are retrieved to form a data row. Then, its five data values ​​at timestamp T2 (220.3V, 15.8A, 0.98, 25.9°C, and 15.5m) are retrieved to form the next row. The same operation is performed on feeder F02. The result is a structured data table with time as the vertical sequence and different sensor data as the horizontal fields. Each row represents a complete status snapshot of a specific feeder at a specific point in time, generating a feeder time series dataset.

[0056] S102: Based on the feeder time series data set, arrange each feeder data group in ascending order by timestamp, extract power values ​​at adjacent time points to form a difference sequence, set a power fluctuation threshold range, identify the starting time points of the sections that continuously exceed the threshold, and generate a synchronous fluctuation time mark group;

[0057] Based on the feeder time series data set, the power value of each feeder at each time stamp is first calculated. The power value P is obtained by the following formula: , where U is the effective value of voltage, I is the effective value of current, The power factor is calculated by taking feeder F01 at time T1 as an example and substituting the data into the calculation:

[0058] P(F01,T1)=220.1V*15.2A*0.98=3277.5W;

[0059] Likewise:

[0060] P(F01,T2)=220.3V*15.8A*0.98=3411.0W;

[0061] P(F02,T1)=219.9V*18.1A*0.99=3937.6W;

[0062] P(F02,T2)=220.0V*18.3A*0.99=3985.4W;

[0063] Then, the data of each feeder are arranged in ascending order according to the time stamps T1, T2, ..., Tn, and the power values ​​of two adjacent time points are extracted to form a power difference sequence. , which is calculated as For example, the power difference of feeder F01 at time T2 is , the power difference of feeder F02 at time T2 is , then set a fluctuation threshold to determine whether the power fluctuation is severe The threshold is set based on the average power of the feeder in one natural day. Standard deviation of power value The specific setting formula is , assuming that the feeder F01 is calculated by retrieving historical data 3100W, is 50W, then its fluctuation threshold , assuming that feeder F02 3800W, is 60W, then its fluctuation threshold Then, the feeder's With their respective , if there is a time point , at which point in time there are more than a preset number (for example, 2) of feeders that meet the Condition, and this state at the next time point If it still continues, then the starting time point Record it, and assume that at a subsequent time point T5, it is calculated that , , and at time T6 both still exceed their respective thresholds, then time T5 is identified as the starting point of a synchronous fluctuation, and a synchronous fluctuation time mark group is generated.

[0064] S103: Calling the synchronous fluctuation time mark group, intercepting the power data sequence of the corresponding time segment of each feeder, calculating the linear regression slope of the power value in each time segment, and concatenating the segments whose absolute values ​​of the slope exceed the set reference value in chronological order to generate the feeder power fluctuation sequence result;

[0065] Call the generated synchronous fluctuation time stamp group, assuming that it contains a starting time point T5, and determine the duration of the fluctuation segment. This duration is retrieved from T5 and backward until the first time point T8 is found. At T8, the power difference of all involved feeders (F01 and F02) is Both fell back to their respective fluctuation thresholds The fluctuation section is determined as [T5, T8], where T5 = 10:01:00, T6 = 10:01:05, T7 = 10:01:10, and T8 = 10:01:15. The power data sequence of feeder F01 in this [T5, T8] section is intercepted, and its power values ​​are assumed to be:

[0066] P(F01,T5)=3500W;

[0067] P(F01,T6)=3820W;

[0068] P(F01,T7)=4150W;

[0069] P(F01,T8)=4500W;

[0070] Calculate the linear regression slope of the power value in this segment, quantify the time point sequence as the independent variable x, that is, x={1,2,3,4}, and the power value sequence as the dependent variable y, that is, y={3500,3820,4150,4500}. The calculation formula of the slope m is: , where n is the number of data points, here n=4, substitute the data for calculation:

[0071] , ;

[0072] ;

[0073] ;

[0074] Substituting into the formula we get , the slope means that the power increases by 333W on average every 5-second time step, setting a slope benchmark value The reference value is set based on the technical specifications of the power supply unit (PSU) of the mounted server equipment. Usually, the power change rate is not allowed to exceed a specific value. Assuming that the value is 100W per second, considering that the time step in this example is 5 seconds, the slope reference value is , the calculated absolute value of the slope and benchmark values For comparison, because Therefore, the power fluctuation of F01 in the [T5, T8] section does not exceed the benchmark, and the data of this section is not selected. If the absolute value of the slope calculated in another section exceeds 500, the complete power data sequence of this section (such as [P(Ts),…,P(Te)]) is taken as an element and arranged in chronological order together with all other section data that meet the conditions to generate the feeder power fluctuation sequence result.

[0075] See also Figure 3 , the steps to obtain the feeder line loss mutation marking result are:

[0076] S201: Calling the power input and output value sequences of adjacent feeder nodes in the feeder power fluctuation sequence result, calculating the input and output differences by aligning the time points, and counting the absolute values ​​of the power differences of each feeder node pair to generate a feeder power difference sequence;

[0077] The feeder power fluctuation sequence results are called. This sequence contains feeder data with rapidly changing power values ​​within a specific time segment. For example, the power fluctuation sequence of feeder F01 in the time segment [T5, T8] is selected, and it is determined that there are three monitoring nodes connected in sequence on the power transmission path of the feeder, denoted as F01-N1, F01-N2, and F01-N3. Among them, F01-N1 is the input node, F01-N2 is the intermediate node, and F01-N3 is the output node, forming two adjacent node pairs (F01-N1, F01-N2) and (F01-N2, F01-N3). The power input value sequence of the F01-N1 node is called from the fluctuation sequence. Its power values ​​at the four time points T5, T6, T7, and T8 are respectively:

[0078] Pin(N1)={3500.0W,3820.0W,4150.0W,4500.0W};

[0079] At the same time, the power input value sequence of the downstream node F01-N2 is obtained. This value is equivalent to the output value of the upstream node pair. The sequence is:

[0080] Pin(N2)={3465.0W,3778.0W,4100.0W,4441.5W};

[0081] And the power input value sequence for F01-N3:

[0082] Pin(N3)={3430.7W,3740.2W,3978.0W,4385.9W};

[0083] These sequences are precisely aligned at time points, and the input and output power difference of the first node pair (F01-N1, F01-N2) at each moment is calculated as follows: ;

[0084] At time T5, the difference is ;

[0085] At time T6, the difference is ;

[0086] At time T7, the difference is ;

[0087] At time T8, the difference is ;

[0088] Perform the same difference calculation for the second node pair (F01-N2, F01-N3):

[0089] At time T5, the difference is ;

[0090] At time T6, the difference is ;

[0091] At time T7, the difference is ;

[0092] At time T8, the difference is ;

[0093] Then, the absolute values ​​of these calculated power difference values ​​are extracted. Since all calculation results are positive, the absolute value is itself. These difference values ​​are organized by node pairs and time sequence. Finally, a structured data containing the power loss situation of each node pair at consecutive time points is obtained, and a feeder power difference sequence is generated.

[0094] S202: extracting the moving average of the difference values ​​in the continuous time window of the feeder power difference sequence, calculating the Pearson correlation coefficient between the increasing trend of the sequence numbers of adjacent feeder nodes and the trend of the difference value change, and generating a trend difference identifier;

[0095] Extract the feeder power difference sequence. Take the difference value sequence {34.3W, 37.8W, 122.0W, 55.6W} of the node pair (F01-N2, F01-N3) as an example. Set a continuous time window with a width of 3 time points and calculate the moving average of the difference value in the window. For the first window starting at T5, its mean is , for the second window starting at T6, its mean is Subsequently, in order to determine whether the line loss distribution of the entire feeder is uniform at a specific time, the data at time T7 is selected. At this time, the power difference value of the node pair (F01-N1, F01-N2) is 50.0W, and the power difference value of the node pair (F01-N2, F01-N3) is 122.0W. Assuming that the feeder has a subsequent node pair (F01-N3, F01-N4), its power difference value at time T7 is 52.0W. Thus, two sequences for correlation analysis are formed. The first sequence is the increasing sequence of the sequence numbers of adjacent feeder nodes, that is, the sequence number sequence X={1,2,3} representing the first, second, and third segments. The second sequence is the power difference value sequence Y={50.0,122.0,52.0} of the corresponding segments at time T7. The Pearson correlation coefficient r of these two sequences is calculated, and the calculation formula is: , where n is the number of data pairs, here n=3, x is the node number, and y is the power difference value. First, calculate the cumulative value of each item:

[0096] , , ;

[0097] ;

[0098] ;

[0099] Substituting these values ​​into the formula:

[0100] ;

[0101] The calculated correlation coefficient r = 0.024 is used as a quantitative indicator of the trend status of the feeder at time T7 to generate a trend difference identifier.

[0102] S203: Filtering node pairs whose Pearson correlation coefficients are lower than a set threshold based on the trend difference identifier, marking the node with the larger sequence number in the node pair as a path abnormal node, and generating a feeder line loss mutation mark;

[0103] According to the trend difference identifier, that is, the Pearson correlation coefficient r value, a correlation coefficient threshold for judging trend consistency is set The threshold is set based on the historical data analysis results of multiple feeders in healthy operation. The analysis shows that the line loss of healthy feeders is evenly distributed, and its r value is usually stable above 0.8. In order to leave enough judgment margin, the threshold is set to , compare the trend difference identifier r=0.024 of feeder F01 calculated at time T7 in the previous paragraph with this threshold, since , the correlation coefficient value is lower than the set threshold, so the feeder F01 is screened out at time T7 with abnormal line loss distribution. It is necessary to further locate the problem node and retrieve the power difference sequence Y={50.0W,122.0W,52.0W} used to calculate the r value. The values ​​in the sequence are checked and the average value of the sequence is calculated. and standard deviation , find the data point with the largest deviation from the average value, where the difference between 122.0W and the average value is , which is much larger than the difference between other data points and the average value. The difference value of 122.0W corresponds to the second node pair (F01-N2, F01-N3). According to the processing rules, the node with the larger sequence number in this node pair, that is, F01-N3, is marked as a potential path abnormal node. The unique identifier of this node, "F01-N3", is recorded, and a feeder line loss mutation mark is generated.

[0104] See also Figure 4 , the steps to obtain the abnormal path map results of line loss structure are:

[0105] S301: Calling the path abnormal node set in the feeder line loss mutation mark, matching the low-voltage access point coordinates, terminal metering equipment numbers and branch bus connection relationships in the distribution network topology structure, and generating an abnormal node topology association table;

[0106] The feeder line loss mutation mark set is called, which contains node identifiers identified as path anomalies. For example, the set is {"F01-N3", "F02-N5"}. For each abnormal node in the set, a query and match is performed in the pre-built distribution network topology database. The database uses the node identifier as the primary key and stores the physical and electrical connection information of each node. For the node "F01-N3", a matching operation is performed, and the corresponding low-voltage access point three-dimensional geographic coordinates are extracted from the database (longitude: 121.4503, latitude: 31.2215, floor: 15.0 meters). The unique serial number of the terminal metering device directly associated with it is "METER-SN-987654", and its connection relationship on the branch bus is recorded as the upstream node "F01-N2", the downstream node "F01-N3". The upstream node "DP-05-Panel-A" is matched with the same matching process for another node "F02-N5" in the set. The query results in its coordinates (longitude: 121.4511, latitude: 31.2209, floor: 28.0 meters), the associated metering equipment number is "METER-SN-987712", and the connection relationship is the upstream node "F02-N4" and the downstream node "DP-11-Panel-B". These discrete information queried and matched are structured and integrated to create a data entry for each abnormal node. Each entry contains five fields: node identifier, access point coordinates, equipment number, upstream connection point, and downstream connection point. The data entries of all abnormal nodes are combined into a tabular data structure to generate an abnormal node topology association table.

[0107] S302: Based on the abnormal node topology association table, extract the power values ​​of each abnormal node for three consecutive time periods, calculate the absolute value of the difference between the power change rates of adjacent periods, and generate the node power fluctuation gradient;

[0108] Based on the abnormal node topology association table, extract the identifiers of all abnormal nodes, such as "F01-N3", and retrieve the power input values ​​of the node in four consecutive time periods, namely T7, T8, T9, and T10, from the feeder time series data set. Assume that the power value at time T7 is The power value is 3978.0W, which comes from the calculation in the previous step. The power value at time T8 is , the power value at T9 is , the power value at time T10 is , the time interval of each time period For 5 seconds, first calculate the power change rate between adjacent cycles. The change rate of the first time period (T7 to T8) is , the rate of change of the second time period (T8 to T9) , the rate of change of the third time period (T9 to T10) , and then calculate the absolute value of the difference between these continuous change rates, which is the node power fluctuation gradient G, and its calculation formula is: , applied to this example, the fluctuation gradient at time T9 is , the fluctuation gradient at time T10 is ,The series of gradient values ​​calculated for each abnormal node at each time point are stored to generate the node power fluctuation gradient.

[0109] S303: Based on the node power fluctuation gradient, identify the node group where the fluctuation direction has reversed twice in succession, construct the conduction path between the nodes according to the distribution network connection order, and generate a line loss structure abnormal path map;

[0110] According to the node power fluctuation gradient data and the continuous power change rate sequence obtained in the gradient calculation process, for example, the change rate sequence of node "F01-N3" is {81.58, -7.18, 9.0}, the sign of each value in the sequence is judged, and the symbol sequence is {positive, negative, positive}. Check whether this symbol sequence has two consecutive direction reversals, that is, look for the pattern of "{positive, negative, positive}" or "{negative, positive, negative}". The rate symbol sequence of node "F01-N3" {positive, negative, positive} meets this pattern. Therefore, node "F01-N3" is marked as a fluctuation reversal node at time T10. Assume that through the same process analysis, another abnormal node "F01-N7" on the same feeder F01 is also identified as a fluctuation reversal node in the same time period. These identified nodes "F01-N3" and "F01-N7" forms a node group. The abnormal node topology association table is then called to find the electrical connection path between the two nodes. The table records that the downstream of "F01-N3" is "DP-05-Panel-A". Assuming that the topology data query shows that "DP-05-Panel-A" is ultimately connected to the upstream node of "F01-N7", a conduction path is constructed. This path is represented as an ordered list of node identifiers, for example ["F01-N3", "DP-05-Panel-A", "F01-N7"]. All identified conduction paths consisting of wave reversal nodes are collected, and each path is logically connected in a graphical manner, where nodes are vertices of the graph and electrical connections are directed edges. Finally, all these paths are integrated to generate a line loss structure abnormal path map.

[0111] See also Figure 5 , the steps to obtain the three-dimensional structure risk positioning information results are:

[0112] S401: Calling the abnormal path node in the line loss structure abnormal path map, extracting the corresponding feeder topology number and distribution level identifier, matching the geographic coordinate code in the power grid GIS system, and generating the abnormal node spatial attribute table;

[0113] The line loss structure abnormal path map is called. The map contains a conduction path consisting of abnormal nodes ["F01-N3", "DP-05-Panel-A", "F01-N7"]. First, the first abnormal node "F01-N3" in the path is extracted. The feeder topology number "F01" to which it belongs is parsed from the identifier of the node. Then, the pre-stored equipment ledger database is accessed and the distribution level identifier of "F01-N3" is found to be "L2-FP15-RMU03". This identifier indicates that this node is located in the secondary distribution layer. , 15th floor, ring main unit No. 03, then the node identifier "F01-N3" is used as the query keyword to match in the enterprise-level power grid geographic information system (GIS) database. The power grid GIS system pre-binds the physical location of each electrical node with a unique geographic coordinate code. Assume that the geographic coordinate code returned by the GIS system for the node "F01-N3" is "BDG-A-F15-ZNE-012". This code indicates that the device is located in the 15th floor of Building A, East Zone Grid 12. For the second node "DP -05-Panel-A". After parsing, its node identifier has no feeder topology number. The distribution level identifier is "L3-FP15-DB05A", which represents the third-level distribution, the 15th floor, and the phase A of the distribution box No. 05. Its GIS coordinate code is "BDG-A-F15-ZNC-045". The operation is performed on the third node "F01-N7". Its feeder topology number is "F01". The distribution level identifier is "L2-FP20-RMU01", which represents the second-level distribution, the 20th floor, and the ring main unit No. 01. The S coordinate code is "BDG-A-F20-ZNW-003". The three pieces of information extracted for each node, namely the feeder topology number, distribution layer identifier, and geographic coordinate code, are integrated to create a table structure with the node identifier as the primary key. For example, the first row of data contains the node "F01-N3", the feeder topology "F01", the layer identifier "L2-FP15-RMU03", and the coordinate code "BDG-A-F15-ZNE-012". Subsequent rows sequentially enter the information of other nodes on the path to generate the abnormal node spatial attribute table.

[0114] S402: Based on the abnormal node spatial attribute table, the hierarchical structure is divided according to the feeder voltage level, the node geographic coordinates are converted into X / Y / Z axis values ​​in a three-dimensional coordinate system, and a node spatial coordinate mapping set is generated;

[0115] Based on the abnormal node spatial attribute table, all nodes in the table are first divided into hierarchical levels according to their associated feeder voltage levels. According to the feeder topology number "F01", its voltage level is queried in the power grid parameter database. The query result is 380V, which belongs to the low-voltage distribution network. Therefore, all nodes in the path ["F01-N3", "DP-05-Panel-A", "F01-N7"] are divided into the "low voltage" hierarchical structure. Then, the geographic coordinate code of each node is converted into a specific value in the three-dimensional Cartesian coordinate system. This conversion process requires setting A coordinate system origin (0,0,0) is set as the ground edge of the southwest corner of Building A. The main facade of the building is set parallel to the X axis, the depth direction is parallel to the Y axis, and the vertical height is the Z axis. For node "F01-N3", its GIS code "BDG-A-F15-ZNE-012" and level identifier "L2-FP15-RMU03" provide positioning information, among which the floor information "F15" is used to calculate the Z axis coordinate. Assuming that the standard floor height is 4.0 meters and the equipment is installed at a height of 1.5 meters from the ground, its Z axis coordinate is , its area code "ZNE-012" (East District Grid 12) is converted to specific X and Y coordinates by querying the floor plan grid map. Assuming that it is converted to , so the spatial coordinates of node "F01-N3" are (50.5, 12.3, 57.5). Perform the same transformation on node "DP-05-Panel-A", which has the 15th floor and the Z-axis coordinates , whose zone code "ZNC-045" (Central District Grid 45) is converted to , its spatial coordinates are (25.1, 15.8, 57.2), transform the node "F01-N7", its floor is 20, and the Z-axis coordinate is , whose area code "ZNW-003" (Western District Grid 03) is converted to , whose spatial coordinates are (5.2, 22.4, 77.5). These node identifiers and their corresponding three-dimensional coordinate values ​​are organized in the form of key-value pairs to generate a node space coordinate mapping set.

[0116] S403: Based on the node spatial coordinate mapping set, a vector relationship of the connection path between nodes is established in the three-dimensional coordinate system, and the distribution level identifier corresponding to each node is superimposed to generate three-dimensional structural risk location information;

[0117] According to the node space coordinate mapping set, the mapping set contains the specific three-dimensional coordinates of each node on the path, such as the "F01-N3" coordinates , "DP-05-Panel-A" coordinates , and the coordinates "F01-N7" , in the defined three-dimensional coordinate system, establish the vector relationship of the connection path between two adjacent nodes on the path. First, calculate the path segment vector from node "F01-N3" to "DP-05-Panel-A" , which is calculated by subtracting the starting point coordinate from the end point coordinate, that is , this vector indicates that the first path segment extends 25.4 meters in the negative direction of the X axis, extends 3.5 meters in the positive direction of the Y axis, and decreases 0.3 meters in the negative direction of the Z axis. Then calculate the second path segment vector from node "DP-05-Panel-A" to "F01-N7". , , this vector indicates that the path segment extends 19.9 meters in the negative direction of the X axis, 6.6 meters in the positive direction of the Y axis, and climbs 20.3 meters in the positive direction of the Z axis. Subsequently, each path segment vector is superimposed and associated with the distribution level identifier corresponding to its starting point and end point. For the vector , the hierarchical identifier of its starting point "F01-N3" is "L2-FP15-RMU03", and the hierarchical identifier of its end point "DP-05-Panel-A" is "L3-FP15-DB05A". This information is integrated into a structured data object, which contains the start point, end point, three-dimensional coordinates and hierarchical information of the path segment. This operation is performed on all path segments, and they are combined into an ordered list. Finally, a composite information body that describes the precise direction, location and equipment hierarchy of the abnormal path in three-dimensional space is obtained, generating three-dimensional structural risk positioning information.

[0118] See also Figure 6 , the steps to obtain the grid line loss visualization layer results are:

[0119] S501: Call the node coordinates and level identifiers in the three-dimensional structure risk location information, match the corresponding node positions in the feeder path topology diagram, set the red dotted line channel rendering style according to the path connection status, and generate the abnormal path rendering base map;

[0120] The three-dimensional structure risk location information is called, which includes the abnormal conduction path ["F01-N3", "DP-05-Panel-A", "F01-N7"] and the three-dimensional coordinates and distribution level identification of each node. The pre-built feeder path topology diagram displayed in a three-dimensional vector model is loaded. In the three-dimensional model, a spatial index query is performed based on the coordinate value (50.5, 12.3, 57.5) of the node "F01-N3". The geometric object representing the node in the model is located and obtained. The same node position matching operation is performed on the coordinates (25.1, 15.8, 57.2) of "DP-05-Panel-A" and the coordinates (5.2, 22.4, 77.5) of "F01-N7" in the path. After confirming that all nodes on the path are in the topology diagram, the node position matching operation is performed. After finding the corresponding geometric objects in the image, the preset connection relationship between these objects is retrieved, that is, the line segments or curve geometries representing the physical cables or busbars. For each retrieved line segment geometry representing the path connection status, its rendering attribute parameters are modified. Specifically, the RGB value of the color attribute is set to (255,0,0), the line type attribute is set to "dashed line", and the dashed line style parameter is set to an array consisting of two numbers [15,8], where the first number 15 represents that the length of the dashed line segment is 15 pixels, and the second number 8 represents that the interval between the line segments is 8 pixels. At the same time, the line width attribute is set to 4 pixels. This process does not change the original geometric data of the topological structure diagram, but only applies a set of independent visual style rules to the specified path part to generate an abnormal path rendering base map.

[0121] S502: Rendering a base map based on the abnormal path, overlaying the equipment point data of the terminal load unit distribution map, assigning layer identifiers of different shapes according to the spatial positions of the nodes, and generating a composite spatial identification layer;

[0122] Based on the abnormal path rendering base map, call an independent distribution map data file that records the information of all terminal load units. The file contains the point information of each electrical equipment. For example, there is a record in the data file, whose equipment type is "precision air conditioning", equipment number is "AC-15-02", and spatial position coordinates are (26.0, 14.5, 56.0). The equipment type of another record is "server cabinet", equipment number is "RACK-15-E-04", and spatial position coordinates are (50.0, 11.8, 56.0). These equipment point data are superimposed on the abnormal path rendering base map as new layer elements. For each newly superimposed equipment point, according to the value of its equipment type field, it is assigned from a predefined mapping table. A specific three-dimensional geometric figure is used as its layer identifier. The setting rule of the mapping table is: if the device type is "server cabinet", a cube figure with dimensions of (1.0, 0.6, 2.0) is assigned; if the device type is "precision air conditioner", a cylinder figure with dimensions of (1.2, 1.2, 1.8) is assigned; if the device type is "head cabinet", a quadrangular pyramid figure with a bottom surface of a square with a side length of 1.0 meter and a height of 2.2 meters is assigned. According to this rule, a cube is instantiated at the coordinates (50.0, 11.8, 56.0) and a cylinder is instantiated at the coordinates (26.0, 14.5, 56.0). All these geometric figure sets generated according to the rule are composed into a new layer to generate a composite space identification layer.

[0123] S503: Integrate the hierarchical division data of the composite space identification layer and the branch structure hierarchical layer, uniformly adjust the transparency and superposition order of each layer, and generate a power grid line loss visualization layer;

[0124] Integrate the composite space identification layer, which contains the red dotted abnormal path and the specific shape identifiers representing various load devices. At the same time, call a division data describing the hierarchical relationship of the power grid branch structure. This data gives each electrical node a hierarchical identifier such as "L2-FP15-RMU03" or "L3-FP15-DB05A". For each visible element in the composite space identification layer, uniformly adjust its rendering transparency parameters and the superposition order parameters that determine its front and back occlusion relationship. The adjustment is based on the hierarchical division data associated with the element. The specific adjustment rules are set as follows: Set the layer superposition order value Z-Index of the basic structure model representing the building wall, floor slab, etc. to 0, its transparency Alpha value to 0.15, and set the Z-Index of the layer identifier representing the third-level distribution equipment (the hierarchical identifier starts with L3) to 0. The dex value is set to 10, the Alpha value is set to 0.7, the Z-Index of the layer identifier representing the secondary distribution equipment (the layer identifier starts with L2) is set to 20, the Alpha value is set to 0.8, the Z-Index of the red dotted channel representing the abnormal path is set to 50, the Alpha value is set to 0.9, and the Z-Index of the layer identifier representing the terminal load unit directly associated with the abnormal path (such as "RACK-15-E-04" with adjacent coordinates) is set to 51, and the Alpha value is set to 1.0. According to this rule, the system will draw layer by layer in the order of Z-Index value from small to large. The layer with a large value will cover the layer with a small value, and different transparency settings allow the covered layer information to remain visible. Finally, the setting and integration of all layer parameters are completed to generate a power grid line loss visualization layer.

[0125] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for intelligent diagnosis and visualization of power line losses based on multi-source data acquisition, characterized in that: The following steps are involved: S1: Obtain the sensor data of each feeder, group them by number and arrange them in time sequence, detect the synchronization of power fluctuations, extract the change trend, and obtain the feeder power fluctuation sequence results; S2: Call the feeder power fluctuation sequence results, analyze the difference between the power input and output of adjacent feeder nodes, compare the power difference with the node sequence number change trend, mark the nodes with inconsistent fluctuation trends as path anomalies, and obtain the feeder line loss mutation marking results; S3: extracting abnormal nodes in the feeder line loss mutation mark result, calculating the power change rate difference, extracting the alternating fluctuation node group, forming a path closure graph, and generating a line loss structure abnormal path graph result; S4: extracting the topological number and position code of the abnormal node in the abnormal path map of the line loss structure, attaching the node to the spatial coordinate, marking the hierarchical distribution, outputting the position and hierarchical information, and obtaining the three-dimensional structure risk positioning information result; S5: Based on the three-dimensional structure risk location information results, map abnormal nodes to the topological structure diagram, adjust the rendering style, determine the layer identification based on the spatial position, complete the visual expression of nodes and paths, and generate the power grid line loss visualization layer result.

2. The method for intelligent diagnosis and visualization of power grid line losses based on multi-source data acquisition according to claim 1 is characterized in that: The feeder power fluctuation sequence results include voltage data, current data, power factor data, conductor temperature rise data and loop length data; the feeder line loss mutation marking results include adjacent feeder node power input values, adjacent feeder node power output values, power difference change trends, node sequence number change trends and path abnormal nodes; the line loss structure abnormal path map results include path abnormal node sets, low-voltage access points, terminal metering equipment, branch buses and path node connection sequences; the three-dimensional structure risk positioning information results include feeder topology numbers, distribution level identifiers, geographic location codes, spatial coordinate positions and node level information; the power grid line loss visualization layer results include feeder path topology diagrams, terminal load unit distribution diagrams, branch structure level diagrams, node and path visual expressions and layer identification forms.

3. The method for intelligent diagnosis and visualization of power grid line losses based on multi-source data acquisition according to claim 1 is characterized in that: The steps for obtaining the feeder power fluctuation sequence result are: S101: Obtain the voltage, current, power factor, conductor temperature rise, and loop length sensor data corresponding to each feeder number of the distribution unit, establish a data group index based on the feeder number, and horizontally splice different sensor data with the same number based on the acquisition timestamp to generate a feeder time series data set; S102: Based on the feeder time series data set, arrange each feeder data group in ascending order by timestamp, extract power values ​​at adjacent time points to form a difference sequence, set a power fluctuation threshold range, identify the starting time points of sections that continuously exceed the threshold, and generate a synchronous fluctuation time mark group; S103: Call the synchronous fluctuation time mark group, intercept the power data sequence of the corresponding time segment of each feeder, calculate the linear regression slope of the power value in each time segment, and connect the segments whose absolute value of the slope exceeds the set reference value in series in chronological order to generate the feeder power fluctuation sequence result.

4. The method for intelligent diagnosis and visualization of power grid line losses based on multi-source data acquisition according to claim 1 is characterized in that: The steps for obtaining the feeder line loss mutation marking result are as follows: S201: Calling the power input value and output value sequence of adjacent feeder nodes in the feeder power fluctuation sequence result, calculating the input and output differences by aligning the time points, and counting the absolute value of the power difference of each feeder node pair to generate a feeder power difference sequence; S202: extracting a moving average of difference values ​​in a continuous time window in the feeder power difference sequence, calculating a Pearson correlation coefficient between an increasing trend of sequence numbers of adjacent feeder nodes and a changing trend of difference values, and generating a trend difference identifier; S203: Filtering node pairs whose Pearson correlation coefficients are lower than a set threshold according to the trend difference identifier, marking the node with a larger sequence number in the node pair as a path abnormal node, and generating a feeder line loss mutation mark.

5. The method for intelligent diagnosis and visualization of power grid line losses based on multi-source data acquisition according to claim 1 is characterized in that: The steps for obtaining the abnormal path map result of the line loss structure are as follows: S301: Calling the path abnormal node set in the feeder line loss mutation mark, matching the low-voltage access point coordinates, terminal metering equipment numbers and branch bus connection relationships in the distribution network topology structure, and generating an abnormal node topology association table; S302: Based on the abnormal node topology association table, extract the power values ​​of each abnormal node for three consecutive time periods, calculate the absolute value of the difference between the power change rates of adjacent periods, and generate the node power fluctuation gradient; S303: Based on the node power fluctuation gradient, identify the node group whose fluctuation direction reverses twice in succession, construct the conduction path between the nodes according to the connection order of the distribution network, and generate a line loss structure abnormal path map.

6. The method for intelligent diagnosis and visualization of power grid line losses based on multi-source data acquisition according to claim 1, characterized in that: The steps for obtaining the three-dimensional structure risk positioning information result are: S401: Calling the abnormal path nodes in the abnormal line loss structure path map, extracting the corresponding feeder topology number and distribution level identifier, matching the geographic coordinate code in the power grid GIS system, and generating an abnormal node spatial attribute table; S402: Based on the abnormal node spatial attribute table, the hierarchical structure is divided according to the feeder voltage level, the node geographic coordinates are converted into X / Y / Z axis values ​​in a three-dimensional coordinate system, and a node spatial coordinate mapping set is generated; S403: According to the node space coordinate mapping set, a vector relationship of the connection path between nodes is established in a three-dimensional coordinate system, and the distribution level identifier corresponding to each node is superimposed to generate three-dimensional structural risk positioning information.

7. The method for intelligent diagnosis and visualization of power grid line losses based on multi-source data acquisition according to claim 1, characterized in that: The steps for obtaining the grid line loss visualization layer result are as follows: S501: Calling the node coordinates and level identifiers in the three-dimensional structure risk location information, matching the corresponding node positions in the feeder path topology structure diagram, setting the red dotted line channel rendering style according to the path connection status, and generating an abnormal path rendering base map; S502: Rendering a base map based on the abnormal path, superimposing the equipment point data of the terminal load unit distribution map, assigning layer identifiers of different shapes according to the spatial positions of the nodes, and generating a composite spatial identification layer; S503: Integrate the hierarchical division data of the composite space identification layer and the branch structure level map, uniformly adjust the transparency and superposition order of each layer, and generate a power grid line loss visualization layer.

Images